- Open Access
Pathogenic functions of host microbiota
© The Author(s). 2018
- Received: 27 April 2018
- Accepted: 29 August 2018
- Published: 28 September 2018
It is becoming evident that certain features of human microbiota, encoded by distinct autochthonous taxa, promote disease. As a result, borders between the so-called opportunistic pathogens, pathobionts, and commensals are increasingly blurred, and specific targets for manipulating microbiota to improve host health are becoming elusive.
In this study, we focus on the functions of host bacterial communities that have the potential to cause disease, proposing the term “pathogenic function (pathofunction)”. The concept is presented via three distinct examples, namely, the formation of (i) trimethylamine, (ii) secondary bile acids, and (iii) hydrogen sulfide, which represent metabolites of the gut microbiota linked to the development of non-communicable diseases. Using publicly available metagenomic and metatranscriptomic data (n = 2975), we quantified those pathofunctions in health and disease and exposed the key players. Pathofunctions were ubiquitously present with increased abundances in patient groups. Overall, the three pathofunctions were detected at low mean concentrations (< 1% of total bacteria carried respective genes) and encompassed various taxa, including uncultured members.
We outline how this function-centric approach, where all members of a community exhibiting a particular pathofunction are redundant, can contribute to risk assessment and the development of precision treatment directing gut microbiota to increase host health.
- Gut microbiota
- Systems biology
- Risk assessment
- Hydrogen sulfide
Pathogens are classified as bacteria capable of causing host damage via specific virulence factors that encompass production of toxins, features allowing attachment to and invasion of epithelial cells and components essential for their viability . Definitions of pathogens and associated virulence factors have been continuously adjusted over the last decades proposing additional aspects to be considered for pathogenicity such as host physiology, where certain bacteria are only able to cause disease in immunocompromised subjects . Additionally, the term pathobiont was introduced to describe commensal, harmless bacteria that can turn hostile under specific circumstances . Methodological advancements in the last decade enabled detailed insights into whole bacterial assemblages and expanded investigations to the community level introducing the prominent term “dysbiosis” that describes altered community structures of host microbiota associated with disease . Recently, the “germ-organ theory” was introduced suggesting oxygen to be the main driver of dysbiosis that is accompanied by a bloom of facultative anaerobic Proteobacteria . As a result, gut homeostasis is disrupted leading to disease due to dysfunction of the microbial organ. The “pathobiome” concept represents another community-wide approach and encompasses all pathogenic agents integrated within their biotic environment . It is organism-centric describing the collection of potentially pathogenic microorganisms in a given community. In hand with those broader concepts and terms describing bacteria (and whole communities) damaging the host, borders between the so-called commensals, pathobionts, and opportunistic pathogens are increasingly blurred, and specific community-wide targets for manipulating microbiota to improve host health are becoming elusive.
Selected (putative) pathofunctions of gut microbiota and associated diseases
CVD, T2D, kidney disease
CRC, liver cancer
IBD, pouchitis, CRC
IBD, CVD, renal failure
Several conditions (e.g., hepatic encephalopathy)
Branched-chain amino acids
Obesity-associated insulin resistance
In this study, we investigated three distinct pathofunctions, namely, the microbial formation of (i) trimethylamine (TMA), (ii) secondary bile acids deoxycholic and lithocholic acid (DCA/LCA), and (iii) hydrogensulfide (H2S), in order to expose various characteristics of pathofunctions and to outline strategies/challenges for diagnostics and treatment.
TMA is produced from dietary quaternary amines mainly via three distinct enzymatic routes with betaine, choline, and carnitine as substrates. Various distinct taxa are reported to encode respective enzymes [6, 7] highlighting that pathofunctions can exhibit both biochemical (different pathways) and taxonomic redundancies. Host hepatic flavin monooxygenases (FMO) subsequently oxidize absorbed TMA to trimethylamine N-oxide (TMAO) that is associated with atherosclerosis and severe cardiovascular disease  as well as kidney disease . It is postulated that TMAO promotes disease through the formation of foam cells (lipid-laden macrophages), a diminishing of the reverse cholesterol transport from the atherosclerotic plaque , and enhances platelet reactivity . Recent gene-targeted studies ubiquitously detected potential TMA-producing bacteria, primarily belonging to Clostridiales and Enterobacteriaceae, in the gut of human, where they constitute, however, only a minor part of the total community (below 1% in most samples) with key players yet to be isolated [7, 12].
The secondary bile acids DCA and LCA are formed by gut bacteria via the multistep 7α-dehydroxylation from cholic acid and chenodeoxycholic acid, respectively. They promote cancer of the colon and the liver via various cytotoxic effects and immune system modulations [13, 14]. A few intestinal Clostridiales strains capable of 7α-dehydroxylation have been isolated, though data on their abundance in situ and major taxa involved are scarce. LCA and DCA are detected in most humans suggesting that respective bacteria are ubiquitously present .
Anaerobic respiration with sulfate, sulfite, or organosulfonates as terminal electron acceptors is widespread in various ecosystems. It is performed by the members of many distinct taxa from both Eubacteria and Archaea , where the dsrAB-type dissimilatory (bi)sulfite reductase forming sulfide from sulfite is the key enzyme. In the gut, bacteria acting on sulfate or organic sulfur-containing compounds including mucin, taurine, and amino acids are ubiquitously detected at low abundances . Desulfobacterales and Desulfovibrionales, particularly Desulfovibrio and Bilophila (the latter does not reduce sulfate), are the key players using fermentation end products (e.g., short-chain fatty acids) and H2 as electron donors . At excessive concentrations, H2S is a cytotoxic gas associated with inflammatory conditions of the gut epithelium such as ulcerative colitis and pouchitis  as well as colorectal cancer .
Quantification and characterization of pathofunctions
Overview of individual datasets included in this study
Jie et al. 
CVD (n = 218) vs. controls (n = 186)
Karlsson et al. 
CVD (n = 13) vs. controls (n = 12)
Qin et al. 
T2D (n = 182) vs. controls (n = 185)
Forslund et al. 
T2D (n = 75), T1D (n = 31) vs. samples from VI
Karlsson et al. 
T2D (n = 43) vs. controls (n = 53)
Le Chatelier et al. 
Obese (n = 161) vs. controls (n = 109)
Feng et al. 
CRC (n = 46), LA (n = 47) vs. controls (n = 63)
Zeller et al. 
CRC (n = 91), LA (n = 15), SA (n = 27) vs. controls (n = 66)
Vogtmann et al. 
CRC (n = 52) vs. controls (n = 52)
Qin et al. 
Cirrhosis (n = 123) vs. controls (n = 114)
Qin et al. 
UC (n = 21), CD (n = 4) vs. controls (n = 14)
Schirmer et al. 
MTG: UC (n = 78), CD (n = 175) vs. controls (n = 55)
MTX: UC (n = 46), CD (n = 121) vs. controls (n = 11)
Mehta et al. 
MTG and MTX of 78 subjects (4 time points)
CutCD genes previously found in various taxa were concurrently driving the abundance increase of the total pathway in patient groups (Fig. 2). Occasionally, disparate abundance alterations of taxa during disease were detected, such as Clostridium sensu stricto that, against the common trend, decreased in T2D (III) and cirrhotic (X) patients, underlining distinct ecology of individual cutCD-carrying bacteria. Main members of cntAB and grdH carriers, i.e., Escherichia/Shigella, Klebsiella, and Clostridium XIVa, respectively, were governing elevation of other TMA-forming pathways. Potential 7α-dehydroxylating taxa containing baiA-I, particularly the metagenomic species Firmicutes bacterium CAG:103, that recruited > 60% of bai-associated reads trended increased in CRC patients (Fig. 2, Additional file 3). Bai genes previously described in Clostridium XIVa displayed higher levels in CVD and T2D patients compared with healthy control groups. The main dsrAB-containing taxa, Desulfovibirio and Bilophila, showed similar behavior and governed total pathway alterations in patient groups.
Specific taxa-function analyses (see in Additional file 4) suggest that taxonomy-based diagnostic approaches can be useful to estimate the abundances of certain pathofunction-carrying groups such as the two major dsrAB-containing taxa Desulfovibrio and Bilophila and the cntAB-exhibiting Enterobacteriacea Escherichia/Shigella and Klebsiella, where abundances of pathofunction genes linked to those genera correlated with the overall, cumulative abundances of all members of the respective taxa. This was not the case for genera containing cutCD- or baiA-I.
Comparison of metagenomic and metatranscriptomic data (datasets XII and XIII)
Targeted metabolite measurement provides another diagnostic level and has proven very useful in the discovery of pathofunctions  as it circumvents the need to detect the total gene pool of a given pathofunction, which can be challenging for the sequence-based omics techniques, particularly if bacterial carriers or enzymatic pathways have not yet been comprehensively identified. Metabolite measurements will remain indispensable for diagnostics and monitoring processes in the future. In conclusion, it is desirable to apply a combination of techniques targeting distinct levels to fully grasp both the pathofunctional potential and its actual activity together with resolving all individual taxa involved in order to perform accurate diagnostics.
Implications of the pathofunction concept for risk assessment and development of intervention strategies to improve host health
The schema provides a basic guideline for risk assessment that requires adjustments for each pathofunction considering individual features. For instance, presence of TMA and LCA/DCA producers (level 2) does not imply availability of precursor substrates (level 1), because alternative energy/carbon sources are usually available for their growth. In other words, the detection of specific pathofunctions represents a minor risk for host damage unless respective substrates are available as well. This is supported by the metatranscriptomic data where transcripts were detected in fewer samples compared with gene abundance results, and only baiA-I showed a positive correlation between gene abundance and expression (Table 3). In contrast, reduction of sulfur compounds is the main energy conservation process for sulfate-reducing bacteria, and increased abundances are most probably directly coupled to the elevated production of H2S as indicated by gene expression results that correlated with gene abundance data (Table 3). Per definition, enzymes that catalyze the formation of precursors of harmful metabolites such as choline from phosphatidylcholine or sulfate from host mucus are not pathofunctions as their products do not harm the host; however, they can play an important role and may be considered for risk assessment. In case of TMA and secondary bile acids, substrates are usually available at low amounts, yet scenarios providing high precursor supplies such as choline/carnitine rich diets or high fat intake (promoting secretion of bile) are frequently occurring. Thus, diet is a key element, and comprehensive measures on both dietary components and community functions are needed to establish specific links between intake of precursing substrates, abundances, and expression of particular pathofunctions and risk for host damage. However, in practice, considering general nutritional habits for risk assessment might often be more useful. For instance, diets high in protein can promote the formation of various detrimental putrefaction products (if bacteria carrying respective functions are present) , and it makes little sense trying to single out each amino acid (with respective pathofunction(s)) as separate risk factors, because interventions focusing on the reduction of specific amino acids are impracticable. Rather, the overall protein intake could be lowered in individuals that harbor bacteria carrying pathofunctional-specific putrefaction pathways at elevated concentrations in order to attenuate the risk of host damage.
Host physiology can play a crucial role as well where, similar to opportunistic pathogens, certain pathofunctions are only harmful in susceptible hosts, which is exemplified by the formation of branched-chain amino acids that are proposed to contribute to insulin resistance only in obese subjects . Also for TMAO-specific risk assessment, host physiology might be included. Both genetic defects, namely, trimethylaminuria, where FMO activities are absent, and genotypic (gender) differences in the potential to form TMAO, with higher enzyme activities in women compared to men, were described [38, 39].
Treatment can act on any of the outlined levels, yet broad, multilevel interventions such as limiting intake of precursors together with the reduction of nutritional niches of carriers, accompanied by boosting detoxification mechanisms, are probably most successful. Targeting nutrition (level 1) is attractive as it interferes at the initial stages reducing pathofunction activity. Furthermore, dietary precursors provide a common therapeutic target independent of the composition of bacterial carriers. Precision interventions become more difficult if (i) multiple, universal precursors are involved (e.g., formation of ammonia); (ii) substrates are essential for host health (e.g., choline); are (iii) of endogenous origin (primary bile acids); or (iv) do not involve any precursors (e.g., bacterial proteases). As discussed above, broader dietary interventions might often be more realistic. An example provides patients suffering from trimethylaminuria (accumulation of TMA in body fluids), who are advised to avoid specific foods like red meat and eggs in order to limit the intake of dietary precursors for the formation of TMA .
Restraining abundances of pathofunctions and growth of carriers (level 2) can be another intervention goal. Use of antibiotics is only advisable in severe cases, and rather gentle, more focused interventions are desirable, where overall community compositions are not fundamentally altered. Targeting broader groups like Enterobacteriaceae that are associated with several pathofunctions by reducing oxygen influx and electron acceptors for anaerobic respiration could be effective , whereas precision treatment specifically targeting individual carriers represents an attractive, more focused approach. However, the latter becomes particularly challenging if taxonomically diverse communities that occupy various niches in the gut ecosystem are involved. For instance, TMA-producers encompass a myriad of diverse taxa encoding distinct metabolic pathways, where carrier community assemblages can greatly differ between subjects . Individualized interventions adjusted for each community type might be appropriate to narrow the spectrum of targets. Furthermore, stimulating commensals that compete for growth substrates with pathofunction carriers could be effective to restrain carriers, especially if closely related bacteria that lack pathofunctions and display large niche overlaps with carriers are involved. For several key members exhibiting choline lyase (TMA) and genes for DCA/LCA formation, phylogenetically closely related, pathofunctionally inactive strains have been isolated . Administration of an array of such strains along with appropriate substrates for providing a competitive advantage over pathofunction-carrying bacteria might be applied for precision outcompeting of carriers.
Blocking activity of pathofunctions represents another target to avoid host damage (level 3). An elegant, successful therapeutic example is the application of 3,3-dimethyl-1-butanol, a structural analog of choline, which inhibits TMA lyases of gut microbiota . Detoxification mechanisms by autochthonous communities provide additional, appealing targets for treatment. A prominent approach represents “Archaebiotics” that refers to the use of TMA-depleting methanogens converting TMA to DMA . The recently identified iso-bile acid pathway in certain Ruminococci that degrades secondary bile acids LCA/DCA serves as another example demonstrating autochthonous bacteria as potential detoxifiers . However, interventions at this level represent the last resort, where detoxification of harmful metabolites is directly competing with host absorption, and detailed information on detoxification kinetics will be crucial to assess applicability for treatment. Finally, altering host physiology to attenuate pathofunction virulence (e.g., reducing TMAO formation in the liver) or to promote detoxification mechanisms such as increasing the capacity of colonic epithelial cells to oxidize H2S  might represent additional intervention targets.
The opportunity of modulating gut microbiota to promote host health is increasingly recognized, yet mechanisms underlying host-microbiota interactions are still poorly understood and targets for treatment remain largely elusive. Here, we focused on the concept of pathogenic functions of gut microbiota that play a role in non-communicable disorders and provide a guideline that can assist their diagnostics, risk assessment, and the development of treatment strategies. Insights into features of human microbiota damaging the host are in its infancy and the pathofunctional spectrum is largely unexplored. The discovery of new pathofunctions can pose major challenges as a manifestation of the disease often requires long-term exposure, which complicates appropriate experiments using model systems. Nevertheless, Koch’s postulates can be applied to establish a particular function as pathogenic, when initiating or increasing its activity in a suitable host causes disease as convincingly demonstrated for TMA(O) in a mouse model . However, even the three metabolites investigated in this study are not exclusively regarded as being harmful. For instance, DCA plays a role in colonization resistance against Clostridioides difficile , and moderate levels of H2S were ascribed beneficial effects  exemplifying the need for establishing dose-dependent information for accurate risk assessment where host damage might not correlate with pathofunction activity in a linear fashion. Cohort studies applying longitudinal sampling together with technological advancements including multiomics technologies provide encouraging environments for revealing additional pathofunction candidates.
For diagnostics, comprehensive databases encompassing the full taxonomic and biochemical diversity play a central role, and adjusted workflows to capture low abundant features might be required, which explains the limited results related to the three pathofunctions obtained in original studies (Additional file 1). Often, key pathofunction carriers are unknown, even for those presented in this study. Metagenomic species identified based on genome reconstructions from metagenomic data circumvent the need for cultivation and proved useful in this study where they served as key references (Fig. 2, Additional file 3). It is possible to estimate their intestinal niches based on genomic features; however, the ecological understanding of such bacteria will be limited due to inability to perform defined experiments. The need to isolate and cultivate key pathofunction carriers remains eminent.
Complete eradication or blocking of all pathofunctions in a given community is difficult, and rather restraining pathofunction abundance and activity will be in focus in the future. Major tasks will involve quantitative monitoring of long-term exposure dynamics to establish concentration thresholds for risk assessment and for defining successful treatment. Although the so-called “healthy microbiota”, derived from symptom-free subjects, provides a first reference, it is an imperfect benchmark that is vaguely defined and contains a myriad of pathofunctions. In our opinion, reducing pathofunctions will improve host well-being, even in the healthy population, and particularly bears great potential when it comes to increase our lifespan and to promote healthy aging where chronic disorders play a central role.
CutCD, cntAB, and grdH (TMA formation)
References for cutCD and cntAB provided in  were updated (PATRIC genomes, n = 107,042, June 2017). To identify genes encoding the β-subunit of betaine reductases (grdH), the same genomes were screened (hmmsearch, HMMER 3.1b1, hmmer.org) using a hidden Markov model (HMM) constructed from the following protein references based on : 742765.5.peg.3571, 1133568.3.peg.2056, 1125712.3.peg.1676, 999407.4.peg.5417, 1531.8.peg.5368, 712357.3.peg.735, 552395.3.peg.1966, 411465.10.peg.881, 457415.3.peg.2639; sequences were trimmed from the 3-prime end till selenocystein as this part was often lacking in PATRIC sequences. A phylogenetic tree was constructed (FastTree (v. 2.1.8)  using the JTT+CAT model) from all sequences that displayed HMM scores > 100 and ≥ 80% coverage to the model, and distances between the branch tips and the top-scoring sequence were determined using cophenetic.phylo function in R (v. 3.1.2) (package: ape, v. 3.4). A steep HMM score drop was obvious at around 550 that correlated with the increases in phylogenetic distances, and all sequences displaying a score > 500 were considered as true grdH yielding 346 candidates (Additional file 6A). Selected sequences form a clade in the tree separated from sequences encoding distinct functionality (Additional file 6B) in the selenoproteins of the glycine/betaine/sarcosine/d-proline reductase family.
BaiA-I (LCA/DCA formation)
Full-length HMM models were constructed for bai A-I genes using sequences based on Reference  and manual BLAST searches (PATRIC genome IDs: 1505.29, 1505.7, 1232454.3, 500633.7, 553973.6, 411468.9, 658665.3, 658085.3, 1123009.3). All PATRIC genomes were screened, and cutoffs were set after obvious HMM score drops for each gene. Subsequently, all genomes exhibiting ≥ 4 genes in synteny (defined as being separated by ≤ 10 genes based on locus tag) were selected as candidates. Additional manual inspections on NCBI yielded baiA,E,F for genomes 165185.6 and 165186.4 that exhibited only three genes in initial searches. For verification, phylogenetic trees were constructed for all genes (baiA-I) where sequences considered as true bai formed a clade separate from lower-scoring genes that were not considered encoding the functions of interest. Finally, 60 bacteria exhibiting the bai operon were revealed (46 were Clostridium sordellii strains).
DsrAB (H2S formation)
In the database provided by Müller et al. , subunits A and B were split and all sequences displaying > 70% length to the references from Desulfovibrio vulgaris (NC_002937) were subjected to FrameBot analysis (v. 1.2, in default mode , with HMMs derived from FunGene ); all protein sequences were subsequently used in BLAST searches (see below).
CutCD (n = 1) and bai (n = 6) genes derived from metagenomic species available in PATRIC and from reference  were added to the databases (only those found in feces and displaying protein sequence similarities > 70% and < 95% to references were considered). Taxonomic affiliations were based on the RDP taxonomy where 16S rRNA gene sequences of genomes were retrieved and subjected to classification using the RDP classifier  as described previously .
Screening for pathofunctions in metagenomic/transcriptomic datasets
Raw reads of all samples were downloaded from the European Nucleotide Archive (http://www.ebi.ac.uk/ena) and the Sequence Read Archive (https://0-www-ncbi-nlm-nih-gov.brum.beds.ac.uk/sra), quality filtered for an average Q score ≥ 20 and length ≥ 70 using Trimmomatic . Filtered reads were BLASTED (blastx using DIAMOND ) against databases described above, and the top-hitting reference was recorded if the query alignment was ≥ 20 amino acids showing ≥ 70% similarity to references. Three single-copy housekeeping genes encoding 50S ribosomal protein L2 (rplB), recombinase A (recA), and CTP-synthase (pyrG) from all PATRIC genomes were included in BLAST searches . For cutCD, cntAB, grdH, and baiA-I, sequences below the set HMM threshold were included at this stage (as done in ) to avoid the possibility of false-positive counts derived from those related genes. Matching read counts were gene length corrected using the median length of respective reference sequences. For each sample, median counts associated with individual pathofunctions were used to calculate pathofunction abundances relative to mean counts linked to the three housekeeping genes of all PATRIC genomes (representing total genomes in a sample) as performed previously . All genes of a pathway had to be detected for considering a pathway being present (for baiA-I, the cutoff was set at four genes). For TMA, calculations were performed for each pathway separately. Thus, throughout the manuscript, pathofunction “abundance” refers to the percentage of bacteria carrying that function. Metatranscriptomic data are presented relative to the mean expression of the three housekeeping genes. Taxa abundance (and expression levels, respectively) comprising individual pathofunctions are shown on the genus level calculated from the cumulative count data of all genes derived from the same genus in pathofunction reference databases relative to mean counts of the three housekeeping genes of all PATRIC genomes. For taxa not affiliated with a genus such as unclassified Clostridiales bacterium SAMEA3545284 or Firmicutes bacterium CAG:103, strain names are given.
Statistical analyses were performed in R: Spearman correlation (package Hmisc), q values (package fdrtool), Student’s t test (function t.test), logistic regression (function glm) (family = binomial), and area under the receiver-operating characteristic curve (package pROC). Generalized linear mixed-effects models were constructed for each disease (function glmer (family = binomial) from package lme4) using dataset as a random effect. Differences in abundance and expression of pathofunctions were assessed based on generalized linear models (function glm (family = binomial)) using presence/absence data and total counts as offset in order to adjust for lower sequencing depth in metatranscriptomic data (3.97 × 106 ± 1.44 × 105 vs. 2.62 × 106 ± 2.47 × 105 (XII) and 3.74 × 106 ± 8.40 × 104 vs. 2.99 × 106 ± 1.39 × 105 (XIII) (mean ± SE)). FDR-corrected Mann-Whitney U tests were done in QIIME (v. 1.9.1, ). Violin plots and heatmaps (based on log-transformed, abundance data (log(x + 1)) were constructed in R using the packages gplots (v. 2.17.0) and ggplot2 (v. 2.2.1). Networks were visualized in cytoscape (v. 2.3.1, http://cytoscape.org, preferred layout with some modifications) considering correlations (p and q < 0.05, Spearman’s rho ≥ 0.35) that were detected in at least three datasets (n = 12).
We thank everybody who contributed to stimulating discussions on this topic and Michael Beckstette for maintaining the HZI bioinformatics cluster.
This study was supported by the Helmholtz Association’s Initiatives on Personalized Medicine (iMed), Aging and Metabolic Programming (AMPro), and the Centre for Individualized Infection Medicine, Hannover, Germany.
MV and SR contributed to the concept. MV contributed to the bioinformatics. MV, SR, and AK contributed to the data analysis. MV, SR, DHP, and TR wrote the manuscript. All authors read and approved the final manuscript.
Ethics approval and consent to participate
Consent for publication
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Casadevall A, Pirofski L. Host-pathogen interactions: redefining the basic concepts of virulence and pathogenicity. Infect Immun. 1999;67:3703–13.PubMedPubMed CentralGoogle Scholar
- Chow J, Tang H, Mazmanian SK. Pathobionts of the gastrointestinal microbiota and inflammatory disease. Curr Opin Immunol. 2011;23:473–80.View ArticleGoogle Scholar
- Tamboli CP, Neut C, Desreumaux P, Colombel JF. Dysbiosis in inflammatory bowel disease. Gut. 2004;53:1–4.View ArticleGoogle Scholar
- Byndloss MX, Bäumler AJ. The germ-organ theory of non-communicable diseases. Nat Rev Microbiol. 2018;16:103–10.View ArticleGoogle Scholar
- Vayssier-Taussat M, Albina E, Citti C, Cosson J. Shifting the paradigm from pathogens to pathobiome: new concepts in the light of meta-omics. Front Cell Infect Microbiol. 2014;4:1–7.View ArticleGoogle Scholar
- Jameson E, Doxey AC, Airs R, Purdy KJ, Murrell JC, Chen Y. Metagenomic data-mining reveals contrasting microbial populations responsible for trimethylamine formation in human gut and marine ecosystems. Microb Genomics. 2016;2:e000080.Google Scholar
- Rath S, Heidrich B, Pieper DH, Vital M. Uncovering the trimethylamine-producing bacteria of the human gut microbiota. Microbiome. 2017;5:54.View ArticleGoogle Scholar
- Wang Z, Klipfell E, Bennett BJ, Koeth R, Levison BS, Dugar B, et al. Gut flora metabolism of phosphatidylcholine promotes cardiovascular disease. Nature. 2011;472:57–63.View ArticleGoogle Scholar
- Stubbs JR, House JA, Ocque AJ, Zhang S, Johnson C, Kimber C, et al. Serum trimethylamine-N-oxide is elevated in CKD and correlates with coronary atherosclerosis burden. J Am Soc Nephrol. 2015;27:305–13.View ArticleGoogle Scholar
- Koeth R, Wang Z, Levison BS, Buffa J, Org E, Sheehy BT, et al. Intestinal microbiota metabolism of L-carnitine, a nutrient in red meat, promotes atherosclerosis. Nat Med. 2013;19:576–85.View ArticleGoogle Scholar
- Risk T, Zhu W, Gregory JC, Org E, Brown JM, Lusis AJ, et al. Gut microbial metabolite TMAO enhances platelet hyperreactivity and thrombosis risk. Cell. 2016;165:111–24.View ArticleGoogle Scholar
- Borrel G, Mccann A, Deane J, Neto MC, Lynch DB, Brugère JF, et al. Genomics and metagenomics of trimethylamine-utilizing Archaea in the human gut microbiome. ISME J. 2017;11:2059–74.View ArticleGoogle Scholar
- Ajouz H, Mukherji D, Shamseddine A. Secondary bile acids: an underrecognized cause of colon cancer. World J Surg Oncol. 2014;12:164.View ArticleGoogle Scholar
- Ma C, Han M, Heinrich B, Fu Q, Zhang Q, Sandhu M, et al. Gut microbiome – mediated bile acid metabolism regulates liver cancer via NKT cells. Science. 2018;360:1–9.View ArticleGoogle Scholar
- Ridlon JM, Harris SC, Bhowmik S, Kang DJ, Hylemon PB. Consequences of bile salt biotransformations by intestinal bacteria. Gut Microbes. 2016;7:22–39.View ArticleGoogle Scholar
- Müller AL, Kjeldsen KU, Rattei T, Pester M, Loy A. Phylogenetic and environmental diversity of DsrAB-type dissimilatory (bi)sulfite reductases. ISME J. 2015;9:1152–65.View ArticleGoogle Scholar
- Fite A, Macfarlane G, Cummings J, Hopkins M. Identification and quantitation of mucosal and faecal Desulfovibrios using real time polymerase chain reaction. Gut. 2004;53:523–9.View ArticleGoogle Scholar
- Blachier F, Davila A, Mimoun S, Benetti P, Atanasiu C, Andriamihaja M, et al. Luminal sulfide and large intestine mucosa: friend or foe? Amino Acids. 2010;39:335–47.View ArticleGoogle Scholar
- Ohge H, Furne JK, Springfield J, Rothenberger D, Madoff RD, Levitt MD. Association between fecal hydrogen sulfide production and pouchitis. Dis Colon Rectum. 2005;48:469–75.View ArticleGoogle Scholar
- Cemal Y, Patricia GW, Kim H, Cross T-WL, Vermillion K, Carroll T, et al. Race-dependent association of sulfidogenic bacteria with colorectal cancer. Gut. 2017;66:1983–94.View ArticleGoogle Scholar
- Jie Z, Xia H, Zhong S, Feng Q, Li S, Liang S, et al. The gut microbiome in atherosclerotic cardiovascular disease. Nat Commun. 2017;8:845.View ArticleGoogle Scholar
- Karlsson FH, Fåk F, Nookaew I, Tremaroli V, Fagerberg B, Petranovic D, et al. Symptomatic atherosclerosis is associated with an altered gut metagenome. Nat Commun. 2012;3:1245.View ArticleGoogle Scholar
- Qin J, Li Y, Cai Z, Li S, Zhu J, Zhang F, et al. A metagenome-wide association study of gut microbiota in type 2 diabetes. Nature. 2012;490:55–60.View ArticleGoogle Scholar
- Forslund K, Hildebrand F, Nielsen T, Falony G, Le Chatelier E, Sunagawa S, et al. Disentangling type 2 diabetes and metformin treatment signatures in the human gut microbiota. Nature. 2015;528:262–6.View ArticleGoogle Scholar
- Karlsson FH, Tremaroli V, Nookaew I, Bergström G, Behre CJ, Fagerberg B, et al. Gut metagenome in European women with normal, impaired and diabetic glucose control. Nature. 2013;498:99–103.View ArticleGoogle Scholar
- Le Chatelier E, Nielsen T, Qin J, Prifti E, Hildebrand F, Falony G, et al. Richness of human gut microbiome correlates with metabolic markers. Nature. 2013;500:541–6.View ArticleGoogle Scholar
- Feng Q, Liang S, Jia H, Stadlmayr A, Tang L, Lan Z, et al. Gut microbiome development along the colorectal adenoma–carcinoma sequence. Nat Commun. 2015;6:6528.View ArticleGoogle Scholar
- Zeller G, Tap J, Voigt AY, Sunagawa S, Kultima JR, Paul I, et al. Potential of fecal microbiota for early-stage detection of colorectal cancer. Mol Syst Biol. 2014;10:1–18.View ArticleGoogle Scholar
- Vogtmann E, Hua X, Zeller G, Sunagawa S, Voigt AY, Hercog R, et al. Colorectal cancer and the human gut microbiome: reproducibility with whole-genome shotgun sequencing. PLoS One. 2016;11:1–13.View ArticleGoogle Scholar
- Qin N, Yang F, Li A, Prifti E, Chen Y, Shao L, et al. Alterations of the human gut microbiome in liver cirrhosis. Nature. 2014;513:59–64.View ArticleGoogle Scholar
- Qin J, Li R, Raes J, Arumugam M, Burgdorf KS, Manichanh C, et al. A human gut microbial gene catalogue established by metagenomic sequencing. Nature. 2010;464:59–65.View ArticleGoogle Scholar
- Schirmer M, Franzosa EA, Lloyd-price J, Mciver LJ, Schwager R, Poon TW, et al. Dynamics of metatranscription in the inflammatory bowel disease gut microbiome. Nat Microbiol. 2018;3:337–46.View ArticleGoogle Scholar
- Lever M, George PM, Slow S, Bellamy D. Betaine and trimethylamine-N-oxide as predictors of cardiovascular outcomes show different patterns in diabetes mellitus: an observational study. PLoS One. 2014;3:1–19.Google Scholar
- Falony G, Vieira-Silva S, Raes J. Microbiology meets Big Data: the case of gut microbiota–derived trimethylamine. Annu Rev Microbiol. 2015;69:305–21.View ArticleGoogle Scholar
- Tang WHW, Wang Z, Levison BS, Koeth RA, Britt EB, Fu X, et al. Intestinal microbial metabolism of phosphatidylcholine and cardiovascular risk. N Engl J Med. 2013;368:1575–84.View ArticleGoogle Scholar
- Kaur H, Das C, Mande SS. In silico analysis of putrefaction pathways in bacteria and its implication in colorectal cancer. Front Microbiol. 2017;8:1–13.Google Scholar
- Newgard CB, An J, Bain JR, Muehlbauer MJ, Stevens RD, Lien LF, et al. Branched-chain amino acid-related metabolic signature that differentiates obese and lean humans and contributes to insulin resistance. Cell Metab. 2009;9:311–26.View ArticleGoogle Scholar
- Christodoulou J. Trimethylaminuria: an under-recognised and socially debilitating metabolic disorder. J Paediatr Child Health. 2012;48:2011–3.View ArticleGoogle Scholar
- Bennett BJ, Vallim TQDA, Wang Z, Shih DM, Meng Y, Gregory J, et al. Trimethylamine-N-oxide, a metabolite associated with atherosclerosis, exhibits complex genetic and dietary regulation. Cell. 2013;17:49–60.Google Scholar
- Wang Z, Roberts AB, Buffa JA, Levison BS, Zhu W, Org E, et al. Non-lethal inhibition of gut microbial trimethylamine production for the treatment of atherosclerosis. Cell. 2015;163:1585–95.View ArticleGoogle Scholar
- Brugère J, Borrel G, Gaci N, Tottey W, Paul W, Toole O, et al. Archaebiotics. Gut Microbes. 2014;5:5–10.View ArticleGoogle Scholar
- Devlin AS, Fischbach MA. A biosynthetic pathway for a prominent class of microbiota-derived bile acids. Nat Chem Biol. 2015;11:685–90.View ArticleGoogle Scholar
- Buffie CG, Bucci V, Stein RR, McKenney PT, Ling L, Gobourne A, et al. Precision microbiome reconstitution restores bile acid mediated resistance to Clostridium difficile. Nature. 2014;517:205–8.View ArticleGoogle Scholar
- Price MN, Dehal PS, Arkin AP. FastTree 2 - approximately maximum-likelihood trees for large alignments. PLoS One. 2010;5:e9490.View ArticleGoogle Scholar
- Wang Q, Quensen J, Fish J, Lee T. Ecological patterns of nifH genes in four terrestrial climatic zones explored with targeted metagenomics using FrameBot, a new informatics tool. MBio. 2013;4:e00592–13.PubMedPubMed CentralGoogle Scholar
- Fish J, Chai B, Wang Q, Sun Y. FunGene: the functional gene pipeline and repository. Front Microbiol. 2013;4:291.View ArticleGoogle Scholar
- Parks DH, Rinke C, Chuvochina M, Chaumeil P, Woodcroft BJ, Evans PN, et al. Recovery of nearly 8,000 metagenome-assembled genomes substantially expands the tree of life. Nat Microbiol. 2017;2:1533–42.View ArticleGoogle Scholar
- Wang Q, Garrity GM, Tiedje JM, Cole JR. Naıve bayesian classifier for rapid assignment of rRNA sequences into the new bacterial taxonomy. Appl Environ Microbiol. 2007;73:5261–7.View ArticleGoogle Scholar
- Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics. 2014;30:2114–20.View ArticleGoogle Scholar
- Buchfink B, Xie C, Huson DH. Fast and sensitive protein alignment using DIAMOND. Nat Methods. 2015;12:59–60.View ArticleGoogle Scholar
- Vital M, Karch A, Pieper DH. Colonic butyrate-producing communities in humans: an overview using omics data. mSystems. 2017;2:e00130–17.View ArticleGoogle Scholar
- Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK, et al. QIIME allows analysis of high-throughput community sequencing data. Nat Methods. 2010;7:335–6.View ArticleGoogle Scholar
- Mehta RS, Abu-ali GS, Drew DA, Lloyd-price J, Subramanian A, Lochhead P, et al. Stability of the human faecal microbiome in a cohort of adult men. Nat Microbiol. 2018;3:347–55.View ArticleGoogle Scholar
- Nyangale EP, Mottram DS, Gibson GR. Gut microbial activity, implications for health and disease: the potential role of metabolite analysis. J Proteome Res. 2012;11:5573–85.View ArticleGoogle Scholar
- Ferreira RM, Pereira-Marques J, Pinto-Ribeiro I, Costa JL, Carneiro F, Machado JC, et al. Gastric microbial community profiling reveals a dysbiotic cancer-associated microbiota. Gut. 2017;67:226–36.View ArticleGoogle Scholar
- Ferenci P. Brain dysfunction in fulminant hepatic failure. J Hepatol. 1994;21:487–90.View ArticleGoogle Scholar
- Hsiao EY, Mcbride SW, Hsien S, Sharon G, Hyde ER, Mccue T, et al. Microbiota modulate behavioral and physiological abnormalities associated with neurodevelopmental disorders. Cell. 2013;155:1451–63.View ArticleGoogle Scholar
- Guo Z, Zhang J, Wang Z, Ang KY, Huang S. Intestinal microbiota distinguish gout patients from healthy humans. Sci Rep. 2016;6:20602.View ArticleGoogle Scholar
- Steck N, Mueller K, Schemann M, Haller D. Bacterial proteases in IBD and IBS. Gut. 2013;89:25–33.Google Scholar